Time magazine creates persuasion mayhem with salt.  First, they offer a good summary of a Cochrane review (look it up) on the outcomes of salt consumption.

Although lowering dietary salt resulted in a small dip in blood pressure, the researchers found no strong evidence that it reduced rates of death in people with high or normal blood pressure. One study suggested that restricting salt in patients with congestive heart failure could even potentially increase risk of death.

Okay, so a well done review and meta-analysis concludes reducing population consumption of salt has no impact on mortality.  Take your Falling Apples with a sprinkle of sea salt!

But.  In the same article, the Time writer notes:

Still, there is plenty of data — and consensus among experts — that excess dietary salt does affect blood pressure and cardiovascular health.

So.  The best scientific evidence we’ve got from the Cochran review claims no effect, yet there’s plenty of consensus among experts to claim there is an effect.

What journalism might call Point Versus CounterPoint is only what a persuasion theorist would call Biased Versus Objective Processing.  You can certainly find experts who will point to cases that prove salt kills and then try to generalize that reductions in salt consumption at the population level would save lives.  Anyone, without or without those little letters following their name, who reasons like this is not a scientist, but rather merely mortal and in the throes of common sense, human nature, and most particularly, Biased Processing.

What’s the difference between Falling Apples and Change We Cannot Believe In?

Persuasion.

Oh.  And, don’t forget the Bubble!

Every field that aspires to science uses numbers to prosecute its business. If You Can’t Count It, You Can Publish It! So, numbers, particularly in the form of statistical analysis are a crucial part of science. Yet, as I’ve demonstrated numerous times in the Persuasion Blog, some science is mere sophistical statistics, those persuasive presentations of p < .0something, the rhetoric of research. The worse the science the better the persuasion, right?

Of course, this is just one fool’s opinion and he’s cherry picking examples to fit his argument. Show me something other than your sarcasm, Steve.

Okay. How about this demonstration of sophistical statistics.

We related the reluctance to share research data for reanalysis to 1148 statistically significant results reported in 49 papers published in two major psychology journals. We found the reluctance to share data to be associated with weaker evidence (against the null hypothesis of no effect) and a higher prevalence of apparent errors in the reporting of statistical results. The unwillingness to share data was particularly clear when reporting errors had a bearing on statistical significance.

This summary gives it up nicely. Three researchers reanalyzed the published statistics in 49 papers in either the 2005 issues of Journal of Personality and Social Psychology of the Journal of Experimental Psychology: Learning, Memory, and Cognition, two well respected psychology journals. These particular papers were chosen because another research team had contacted the authors of the studies in a previous project, merely asking for a copy of the datasets used in the publications. Some of the 49 authors provided the data, some didn’t. After waiting five years (5 years!!!), the current team pulled the studies and checked the results sections for errors and inconsistencies.

As the researchers noted in the Abstract they found that authors who would not disclose data had more errors of statistical analysis and that the tests of statistical significance were much more likely to be extremely close to the p < .05 level. Here’s a pie chart that displays errors by data shared or not.

Even among researchers who shared the data, there were errors in their analyses, but just eyeballing the differences between the two groups, you can see that folks who refused to share data (after five years!) made more of all kinds of errors. And, the differences are Medium to Large Windowpanes, 35/65 to 25/75 differences, so they are obvious, practical, relevant. What’s more, authors who did not share had data with marginal results; they were more likely to report p values at or near the traditional .05 alpha while authors who shared data found results with much smaller alphas (> .001). Here’s a bar chart to illustrate.

You can see that the gray bars represent authors who did not share and that they had more errors at or near .05 and .01, traditional, almost ritualistic, markers of effect. You can understand why they were reluctant to share.  If you found results, but didn’t share your data, chances were good the results were small effects that you had to finagle to achieve even statistical significance. No wonder these authors found good reasons to withhold their data even after five years of waiting.

Oh, and if you’re not familiar with the publication ethics of publishing in these journals, you need to know that all authors have to sign a contract when they publish stating that they will share data when it is requested. This is not a matter of personal preference or taste; it is a professional standard of behavior with your signature of agreement and consent on it.

Authors who don’t share data are not doing good science. Their inaction violates both the letter and spirit of a contractual agreement they made when publishing. They obviously withhold data because they know they engaged in sophistical statistics and if anyone else ran the data, they’d expose the rhetorical research.

So, through a thoughtful research project on statistical analysis in peer review journals we actually learn a lesson about human nature and persuasion.

All Bad Science Is Persuasive!

Wicherts JM, Bakker M, Molenaar D (2011) Willingness to Share Research Data Is Related to the Strength of the Evidence and the Quality of Reporting of Statistical Results. PLoS ONE 6(11): e26828.

doi:10.1371/journal.pone.0026828

My former office mate, Tim Levine, weighs in on the never ending battle for Better Science with his latest attempt to right the wrongs in journal publication.  Tim’s focus is upon proper thinking with statistics and he provides a nice example ripped from the pages of journal research to illustrate.  Tim’s lesson is easy, fun, and popular, but does require counting, and alas, will probably leave the great unwashed only refreshed, but still filthy.

Tim’s right.  People don’t think correctly with statistics and merely default to the p < .0something as the Eureka moment when there’s more going on.  If only you’d count just a little longer, a little closer, a little harder.

Tim’s most important observation is this:

The null hypothesis is almost never literally true.

Stated another way, you can almost always invent good news even when there is really no news at all.  Null hypotheses are like dead snakes – you still think they could hurt you.  And, when you trip over a rock to avoid the dead snake, you have evidence that snakes are dangerous, even when dead, thus proving dead snakes are a threat which is publishable, but missing the science that fearful people are clumsy.

Tim offers five recommendations that generally counsel patience, replication, and objectivity.  Of course, none of those things will earn promotion or tenure and until we remove that from the equation we’ll continue with the fearful work that populates all journals.  Man, if you cannot break the Laws of Statistical Analysis, you will not publish, but will perish.  You must trip over rocks in your anxiety about the dead snakes in your data; that’s the path to success, but not to science.

Tim Levine (2011).  Statistical Conclusions Validity Basics: Probability and How Type 1 and Type 2 Errors Obscure the Interpretation of Findings in Communication Research Literatures.  Communication Research Reports, 28 (1), 115-119.

DOI:10.1080/08824096.2011.541369

P.S.  You can imagine how well Tim and I got along as office mates, both of us snickering over the foibles, follies, and failures of our colleagues.  It’s a wonder either of us got out of grad school alive.  Mad dogs, Englishmen, and methodologists!

There’s something happenin’ here.
What it is, ain’t exactly clear.

Let’s turn Buffalo Springfield inside out and follow their classic, For What It’s Worth (YouTube), into climate change. The Wall Street Journal has published a signed editorial with sixteen scientists who affirm that Global Warming does not exist and that CO2 poses no threat to the environment.

A candidate for public office in any contemporary democracy may have to consider what, if anything, to do about “global warming.” Candidates should understand that the oft-repeated claim that nearly all scientists demand that something dramatic be done to stop global warming is not true. In fact, a large and growing number of distinguished scientists and engineers do not agree that drastic actions on global warming are needed.

The sixteen signers then note the risks for public disagreement with Global Warming advocates. They cite one outstanding case.

In 2003, Dr. Chris de Freitas, the editor of the journal Climate Research, dared to publish a peer-reviewed article with the politically incorrect (but factually correct) conclusion that the recent warming is not unusual in the context of climate changes over the past thousand years. The international warming establishment quickly mounted a determined campaign to have Dr. de Freitas removed from his editorial job and fired from his university position. Fortunately, Dr. de Freitas was able to keep his university job.

Of course, Buffalo Springfield anticipated this outcome when they sang.

Paranoia strikes deep,
Into your heart it will creep.
It starts when you’re always afraid,
Step out of line and the man comes and takes you away.

This may be the largest public retort to the scientific consensus claim from Global Warming advocates. While there have always been scientists who disputed those advocate claims, they tended to do science rather than persuasion and felt no need to sign silly petitions as if they were voting on gravity. Now, at least 16 are willing to make a high profile persuasion play in public.

Even still, the Buffalo Springfield lyric hits it.

There’s something happenin’ here.
What it is, ain’t exactly clear.

Something is happening both in the science and persuasion of climate change, but it ain’t exactly clear. This editorial is a terrible challenge to the Scientific Consensus Cue so beloved of advocates. They now must devolve into a credentials swamp, shouting My Experts Are Experts and Your Experts Aren’t! an argument no citizen wishes to hear. When advocates are fighting over CVs, they have lost whether they realize it or not.

The safest persuasion play for advocates is Silence. Don’t even acknowledge the signed editorial exists. Just keep flowing on the great wave of Truth. Ignore that Other Guy Behind the Curtain. Persist with the Al Gore PowerPoint Show and all those confident claims of Scientific Consensus. This editorial and each contrary voice changes nothing.

That’s the persuasion play. And, best of all, it requires no science!

Stop, children, what’s that sound,
Everybody look what’s going down.

If you are Weatherman 2.0, you know that things aren’t going your way even with a congenial President and Senate and those two golden, but wasted, years of 2009 and 2010.  The Weather is still getting worse, but worser still no on is listening to you.  That email foolishness.  Al Gore’s Green fortune.  All those lamer scientists publicly disagreeing with the Scientific Consensus.  And, now, come to find out that the Weather Channel is killing you, too.

Joslyn, Savelli, & Nadav-Greenberg present an illuminating four  study package that demonstrates how weather forecasts generate mistrust and confusion in consumers through an emphasis upon Worst Case reporting.  The problem stems from well known anchoring effects.  Let’s read their conclusion.

These four studies extend the well-known anchoring effect to worst-case scenario forecasts informing weather-related decisions. It is clear that people’s understanding of future weather conditions is influenced by the kind of uncertainty information provided in the forecast. Participants with the worst-case scenario had a biased understanding despite having the same single-value forecast as those with other formats. Remarkably, the anchoring effect was observed when the low likelihood of the anchor was clearly specified and when a better standard was available . . . Many forecast providers believe that the worst-case scenario is the uncertainty information most understandable and beneficial to general public end users.  Indeed, the worst-case scenario may well reduce the amount of information people must process by focusing on values that are critical to the decision at hand. However, the research reported here suggests that this reduction in uncertainty information can seriously mislead the user. It appears to convince people that wind speeds will be higher and temperatures will be lower than what are indicated in the forecast.  One may be tempted to think that such a bias is advantageous in potentially dangerous situations because it could convince otherwise reluctant people to take precautionary action . . . The research reported here suggests that the worst-case scenario leads to a misunderstanding of the forecast, which could have serious long-range impact.

While Joslyn et al. focus literally upon the daily weather forecast and forecasts for extreme events like hurricanes, the implications for the Global Warming Wars should be apparent, if only to me.  People think about the information they get and how that information is presented can produce unintended consequences.  If you take the time to read and understand this research, you realize the larger implications for the Climate Change Chorus and how their ineffective communication has played a major and perverse role in making things worse for their own case.  The emphasis upon that Worst Case Scenario is the go-to Frame (or Meme or Narrative or Whatever) with a simplistic emphasis upon a single point – average temperature – rather than boundaries or limits or ranges.

Stated another way, the persuasive communication from the Climate Change Chorus and the emphasis upon the Worst Case and the single point estimate, generates biased thinking from people that actually reduces the believability and effectiveness of the communication.  The more the CCC talks, the worse it gets.  Thus, even if science proves that human activity has caused climate change and that climate change will produce negative outcomes for people and (last one!) reducing carbon consumption will make things right, then the CCC communication makes that better world less likely through their pounding and relentless emphasis upon the Worst-Case Scenario.  In so doing, the CCC artlessly creates an anchor, a standard of comparison, a line of judgment, that guarantees less concern, worry, or risk.

Hear a different Chorus?  It’s Greek.  And ancient.  It’s chanting:  Killing what you love.  Killing what you love.  Killing what you love.

Mavens of all Rainbows, learn from failure.  Even if the sky is falling, shouting that The Sky Is Falling may only ensure that everyone goes to the beach.  When its gonna rain anchors, people use a different umbrella.

Joslyn, S., Savelli, S., & Nadav-Greenberg, L. (2011). Reducing probabilistic weather forecasts to the worst-case scenario: Anchoring effects. Journal of Experimental Psychology: Applied, 17(4), 342-353.

doi:10.1037/a0025901

P.S.  Jeepers.  See any applications with warning labels and their Single Point Estimates for Worst Case Scenarios?  All Bad Persuasion Is Sincere, kids.  Just because it feels so right to you doesn’t mean you can change the Other Guys with it.  No wonder only zealots persuade like this.

P.P.S.  Of course, if the CCC stopped with the Worst Case Scenario and Single Point Estimates and switched to Boundaries and Ranges, it might make their case even worse.  They’ve already admitted that the presumed increase in temperatures isn’t statistically significant which is another way of saying Nature is in her Container.